This would honestly work. It forces the attacker to go through with the clearing phase which simultaneously makes it possible to "cancel" the TX through another logic branch before the timeout occurs. I'd say that would meet the needs of a clearing mechanism / fraud prevention system for an exchange perfectly while requiring minimal changes to the software.

Very, very smart idea. A++, would read again.

On Thu, Aug 4, 2016 at 9:55 AM, Tier Nolan via bitcoin-dev <bitcoin-dev@lists.linuxfoundation.org> wrote:
On Wed, Aug 3, 2016 at 7:16 PM, Matthew Roberts via bitcoin-dev <bitcoin-dev@lists.linuxfoundation.org> wrote:
The reason why I bring this up is existing OP codes and TX types don't seem suitable for a secure clearing mechanism;

I think reversing transactions is not likely to be acceptable.  You could add an opcode that requires that an output be set to something.

[target script] SPENDTO

This would require that [target script] is the script for the corresponding output.  This is a purely local check. 

For example, if SPENDTO executes as part of the script for input 3, then it checks that output 3 uses the given script as its scriptPubKey.  The value of input 3 and output 3 would have to be the same too.

This allows check sequence verify to be used to lock the spending script for a while.  This doesn't allow reversal, but would give a 24 hour window where the spenders can reverse the transaction.

[IF <1 day> CSV DROP <live public key> CHECKSIG ELSE <offline protected key> CHECKSIG] SPENDTO <live public key2> CHECKSIG

Someone with the live public key can create a transaction that spends the funds to the script in the square brackets.

Once that transaction hits the blockchain, then someone with the <offline protected key> has 24 hours to spend the output before the person with the live keys can send the funds onward.

_______________________________________________
bitcoin-dev mailing list
bitcoin-dev@lists.linuxfoundation.org
https://lists.linuxfoundation.org/mailman/listinfo/bitcoin-dev